Hacker News Reader

Comment by cromka | Hacker News Reader

The radio limits are very heavily guarded by the firmware solely because its regulatory compliance: different jurisdictions have different limits and you can’t just let userspace or kernel change it at will. Heck, most manufacturers don’t let you change WiFi card anymore or allow only a whitelisted WWAN card for same reason.

So, I would be super interested how you inferred it was Linux’ fault but I sssume b/s at this point.

View on HN

Well, I'd say it's Asus fault for having crappy firmware - but Linux triggers it - causing the card to permanently no longer come up on boot.

"Fried" was a poor choice of words - the device is dead - but it's due to firmware and probably something to do with sleep states.

I didn't mean to imply the radio was actually fried.

It'll be interesting to see what the technicians end up doing to get it working. It might be enough to disconnect the card and force some even deeper reset than a power cycle with drained laptop battery does - but as I still have valid guarantee on the machine, I stopped short of opening it up. Especially since I might need a new wifi card anyway.

If I were to wild-guess here, it sounds like and rfkill in nvram that isn't properly handled by the Asus EC driver for your laptop model. This is a fairly common issue for Windows-first laptops and Asus EC drivers are reverse engineered because they don't support Linux themselves. I assume switching to Windows was not bringing it back anymore?

Chance is you could probably have Claude or Codex had look into it and have them fix it for you.

I had ai look into it, and that took it from works in windows, shows up in Linux - to dead and gone from both os'.

Ed: user AI on the Linux side, to help with getting firmware blobs etc - might be random combination of flaky firmware, flaky Linux support and reboots that did it - not AI or any particular driver/firmware version.

It's the first time I've seen drivers knock out a hw component completely on a consumer product - but I gather the platform is fickle.

The arm64 Snapdragon elite support is very poor unfortunately (due to there being zero upstream support).

Which wouldn't be so bad if they weren't marketed with a promise of Linux support.

Well, thanks for the write up. Will be interesting to hear what they fixed, but chances are they'll just replace the board and not think twice about it.

Reply on news.ycombinator.com